AM1050 Mix Sequencer

ARP 1050 Module

Overview This 5U module is a replica of the 1050 Mix Sequencer module from the ARP 2500 synthesizer. This is a straight forward analog switch but with some rather cool features. The original was designed back in December 1970 using TTL logic chips and FET transistors to switch the signals off and on. This module pre-dates the arrival of CMOS logic, analog switch chips, LED’s and even decent Op Amps. Pearlman was pushing the limits of late 1960’s analog electronics!

The 1050 is one of the modules that puts the 2500 into a new dimension of musical creativity, along with the ModAmp, Dual S&H and UAF. I am building two for my replica.

Description This module has sequential switching for the rapid selection of preset waveforms and signals into a conventional audio mixer format. There are two four-input mixers with electronically gated inputs, an eight-step counter and clock, and associated logic and switching circuitry. Each input has its own attenuator and each output has a master gain control.

The module can be used as an eight input mixer with two adjustable outputs or as two separate four input mixers. A column of illuminated push buttons indicates which inputs are gated on. The switches are pushed to change the state of an input, push-on/push-off. A column of “Exclusive-on” push buttons will turn on a particular input while simultaneously turning all the others off.

ARP 1050 PCB

Original Circuit The 1050 uses (now obsolete) TTL logic chips for the digital control of an audio mixer, with a 74155 dual 2 of 4 decoder driven by a 4-bit binary counter 7493 chip, and clocked by a internal analog pulse generator that uses a MU4894 uni-junction transistor at its core (I have used a 2N4870).

The voltage controlled clock and the TTL logic provides the eight step counting. A set of 7474 flip-flops provide the manual ON/OFF latching of the square push buttons, as well as the XOR function. Four 75451 peripheral driver chips light the incandescent 12V lamps and drive a set of eight discrete FET analog switches.

Additional logic provides individual control of the analog switches from another ARP 2500 module, such as the 1027, via an internal 9 pin RS232 cable. The external signals are buffered by two Motorola MC9818 hex inverter chips which are early 1960’s RTL (Resistor Transistor Logic), typically used in mainframe computers (and the Apollo Guidance Computer) before TTL logic was launched in 1965.

Loads of wires!

Inside each gate is a couple of resistors and a transistor, and they have a +3.6V power rail. In the 1050 they are powered from +5V and switched OFF/ON at a chip level by connecting the chip ground pins to either +5V or digital ground by a front panel switch.

RTL Inverter

ARP 1050 Chips

The picture shows a later 1050 with logic chips from around 1973, although it looks like one has been repaired with an old Sprague 7474 chip from 1968! Sprague were better known for tantalum capacitors (which they invented in WWII) and they briefly entered the TTL market in 1966. They set up a silicon fab facility not far from ARP in the Boston area but they failed to grow this business.

The eight audio signals go via level potentiometers and the FET switches before they are mixed into the 2 separate audio outputs using four 1339 Op Amps. The mode switch can either mix all 8 channels into the 2 outputs or separate 1-4 and 5-8 into two channels. The 1339 have offset trimmers working from a +0.8V power rail. I have omitted this circuit as I am using modern Op Amps with low offset and fast slew rates.

The 1050 contains a huge amount of manual wiring from the front and rear of the PB and it must have been very time consuming to build and test.

The AM1050 I obtained a full set of paper blueprints and schematics in 2006, and transcribed the circuit into Eagle CAD. The ARP circuit requires a PCB board of 180 x 180 mm, and is too large for my 5U system. I wanted to modernise some of the circuit, by using modern analog switch chips (like the SSM2402 and DG442), which also saves space. Initially I laid the circuits onto two PCB’s:

  • Digital PCB with all the digital logic and voltage controlled clock.
  • Analog PCB with analog switches, Op Amps and Audio Outputs.

The original circuit has been retained with the following planned changes:

  • LS TTL logic to reduce power requirements.
  • Click-less analog switch chips to replace the discrete FET’s.
  • Upgrade the digital input.
  • Modern audio Op Amps.

Front Panel The panel is 5U MOTM format, but with the jacks mounted at the base of the panel. There are a couple of “super” toggle switches to source; a 4 pole DPDT and a 3 pole DPDT with momentary in one direction. These are still available but expensive at £10-20 each. The Switchcraft square illuminated buttons and the red round switches are still available from Mouser and I have used ARP style control knobs.

Controls There are 8x 100k Log 270X potentiometers on the input side of the mixer and 2x PRV6 10k Log output level potentiometers. There is a clock frequency (Pulse Rate) control on a PRV6 10K pot, along with an 8x position switch to control the sequence length and to switch the sequencer OFF. There is a 4PDT toggle switch (C&K 7411) to control the mode of the sequencer and another toggle switch switches the clock on or off or enables manual advance. There is also a toggle switch to enable/disable external control of the 1050, which simply pulls the power off the input logic chips (MC9818), a slightly crude method! The clock frequency has to be trimmed at setup time, the highest speed is 33 Hz.

Connections The 1050 has the following external connections:

  • 8x Analog Inputs
  • Clock Frequency Modulation
  • External Advance
  • Analog Outputs A and B

Build History & Outcomes With renewed energy in the summer of 2015 I carefully laid out the two PCB’s. The main digital PCB was tested in September and after a few issues it was nicely sequencing the lights. However the momentary push button response was poor, and I made changes to the voltage levels that it switches, so it drives the 74LS74 chip correctly. In October I had a panel manufactured and fitted the digital board.

In December 2015 I wired up all of the digital board and switches and ordered the Analog PCB, however whilst the clock worked the switching of the channels was error prone and the whole lot locked up. This was the only issue, aside from the clock range needing adjustment. The project was paused until the Spring of 2019 when I swapped in a set of original TTL chips (not LS) with some slight concern as the power consumption had already reached 100 mA with TTL LS. The TTL chips revealed two problems:

  • The 7474 chips were not edge triggering to an OFF state.
  • Complete lockup of the 155 chip with all outputs ON.

In lock up the power consumption ran up to nearly 500 mA! I continued to work on the issues, but a software version looked a better option to ensure 100% operation and lower power consumption. Looking at some 2500 videos you can see the same edge triggering problem on actual 1050’s, with users trying 2 or 3 times before the switch latches.

Software Version In spring of 2018 I started working on a software version using the PIC18F46K22, which has sufficient I/O pins to control most of the logic, except the external input and control which is done in CMOS hardware. Some use of BCD encoding and decoding is needed to keep within the 40 pins. The analog design remains the same using SSM2402 analog switches and high quality Op Amps. The hybrid design means this can all fit on a single PCB. I continued the development in 2019 and moved to a 44 pin SMD version to get further I/O pins. I developed the pseudo code in March but the project was cancelled.

The 7474 Issue The original TTL SN7474 chip was introduced back in 1966 and was available for US$10 when the 2500 was designed in 1970. It is a positive edge triggered D Flip Flop which is frequently used in this application of latching a digital signal with a momentary switch. To make this work effectively the switch needs debouncing, so that when the contacts are settling from a change of ON/OFF state they don’t re-trigger.

The ARP design uses a simple 100R and 100nF capacitor to debounce each channel, with the switch going high when pushed. The high voltage is set by a 470K resistor to +5V and a 100nF capacitor and 2R7 resistor to ground and this circuit is shared across all channels. The trigger signals into the 7474’s are not properly debounced and whilst they sometimes reach the TTL threshold voltage of 2.0V this is not consistent and there is cross talk to other switches.

A modern approach is to use a Schmitt trigger ahead of the flip flop circuit, either in built (Philips use this approach in the PC74HCT74) or as a separate chip such as the MC14490 or MAX6818. I tested a MC14490 debounce chip in front of the 7474 flip flop chips on 8 April and this worked perfectly, with clear ON and OFF states.

I also tested the PC74HCT74 chip which is available as a DIL-14 and can therefore be used in the original ARP 1050. I tested the SMD MAX6818 octal debounce chip, which has a small footprint and is cost effective at around 5 GBP. The chip is mounted on a Switch PCB with the illuminated push buttons, making wiring a lot easier.

External Input The ARP design uses the MC9818 RTL chip to buffer between the 10V signals generated by the 1027 and the 5V TTL logic in the 1050. It is also used to enable or disable this input by switching the chip power pins to +5V or GND. The MC9818 is obsolete and expensive to buy, so I want to design a cheap solution which translates the voltage down. I went for replicating the RTL logic with a 8x simple NPN transistor buffers and retained the 18K input resistors to limit the current.

B1050 PCB rear

2019 Update I resolved the counter lock up issue by using a separate isolated 12V power supply to light the bulbs in the illuminated switches. The REV02 PCB has an isolated 12V power supply for the bulbs so they do not interfere with the digital logic.  The voltage controlled clock was calibrated and runs from one pulse every 4 seconds to over 100Hz, the original goes as low as one pulse every 30 seconds.

I added a new PCB to mount the eight audio input potentiometers to make the build simpler with less wiring. Further progress by sourcing the correct 4PDT toggle switch (C&K 7411) and wiring this into the module, so I can select the different modes, and fitting the 3 pole 3 way toggle switch to the panel which selects OFF/ON/ADVANCE.

Behringer Version Whilst developing the Behringer 1050 in 2020 I designed a CMOS version, which has proved very successful as a commercial product. The CMOS version has to invert the logic states from negative (TTL) to positive (CMOS). I have bought one, and written a blog post about the module here.

2024 Update The 2500 Rack Project was closed in 2020 but I would still like to complete two 5U 1050’s, as I have the panels and switches. The plan is to use the CMOS circuit I designed for the Behringer 1050, with a PCB mounted parallel to the 5U panel.  A test circuit of the XOR push button design was completed in May 2024 for the AM1026, and this proved the switch logic for both modules.

The next step is to lay out the main sequencer logic with analog switches replacing the JFET’s in the original.



Copyright AMSynths 2024